Does omitting mandatory commas affect the reading process?

Talk presented at ECEM 2022, Leicester

Bernhard Angele¹, Ismael Gutiérrez Cordero², Manuel Perea³ and Ana Marcet³

1. Bournemouth University, UK; 2. University of Málaga; 3. University of Valencia

Introduction

What does a comma do

What does a comma do. I have refused them so often and left them out so much and did without them so continually that I have come finally to be indifferent to them. I do not now care whether you put them in or not but for a long time I felt very definitely about them and would have nothing to do with them. As I say commas are servile and they have no life of their own,and their use is not a use, it is a way of replacing one’s own interest and I do decidedly like to like my own interest my own interest in what I am doing. A comma by helping you along holding your coat for you and putting on your shoes keeps you from living your life as actively as you should lead it and to me for many years and I still do feel that way about it only now I do not pay as much attention to them, the use of them was positively degrading.

Gertrude Stein, Lectures in America (1935)

Gertrude Stein (1934). Carl Van Vechten, Public domain

What does a comma do?

Two main purposes:

  1. Signal the “covert prosody” of written language (Chafe, 1988)

  2. Indicate the syntactic structure of a sentence

History of punctuation rules

  • Punctuation did not exist for much of the history of writing.
    • Speakers used individual punctuation systems to help them read aloud
    • With the advent of silent reading, this then evolved into a more universal system to clarify the intended syntactical structure of a sentence
  • As the role of writing broadened from serving mainly in the production formal and carefully redacted documents to everyday individual expression (Figueras Bates, 2015), the role of commas has changed again, with writers adopting more idiosyncratic punctuation strategies in informal communication.

Punctuation is inconsistent

  • Writers have always disagreed as to where they think commas are appropriate, but with text messages becoming more popular, these idiosyncrasies are magnified
  • For example, writers differ in whether they apply the serial (Oxford) comma, as described e.g. by Truss (2003)
  • Some authors are very prescriptive about the use of commas
  • But does “standardised” comma usage actually benefit the reading process?

Lynne Truss: Eats, Shoots, & Leaves

Evidence for the benefits of punctuation

  • Hirotani et al. (2006) found that using non-mandatory commas seemed to facilitate overall reading compared to omitting commas, although there were higher dwell times ahead of the commas.

  • There is not much else information about the effect of commas – mandatory or not –on eye movements

Commas in Spanish

  • In Spanish, most commas are strictly mandatory according to prescriptive grammars. We can distinguish between six uses of commas:
  • Concessive
  • Adversative
  • Connective
  • Enumerative
  • Incisive

Concessive commas

Paula no podrá irse de la clase, aunque acabe el examen pronto.

Paula cannot leave the classroom, even if she finishes the exam soon.

Critical areas:

  • Comma

  • Post-comma 1

  • Post-comma 2

Adversative commas

Quiero contarles una cosa importante a mis abuelos, aunque no lo voy a hacer.

I want to tell an important thing to my grandparents, although I won’t do it.

Critical areas:

  • Comma

  • Post-comma 1

  • Post-comma 2

Connective commas

Siempre se le veía entrenando y, finalmente, consiguió ganar el campeonato.

We always watched him practicing and, finally, he managed to win the championship.

Critical areas:

  • Comma 1

  • Comma 2

  • Post-comma

Enumerative commas

Este año hemos viajado a Bruselas, Lanzarote y Toledo y nos ha gustado mucho.

This year we have traveled to Brussels, Lanzarote and Toledo and we have liked it a lot.

Critical areas:

  • Element 1 (comma)

  • Element 2

  • Element 3 (and)

Incisive commas

Mis dos tortugas, Paula y Alba, tienen más de siete años.

My two turtles, Paula and Alba, are more than seven years old.

Critical areas:

  • Pre-incision (Comma 1)

  • Incision (Comma 2)

  • Post-incision

Hypotheses

  • If we ask participants to read sentences of these types with the commas omitted:

    • If mandatory commas are indeed critical for comprehension, we should see longer overall sentence reading times compared to a control condition with the commas included.
    • In word-based fixation time measures, we should also find evidence of wrap-up effects on the first critical area when the comma was present (and possibly the second area when there was a second comma, i.e. in the case of connective and incisive commas).

Hypotheses (continued)

  • If mandatory commas are optional and do not confer a benefit to experienced readers, we should see no difference in overall sentence reading times between no comma and control.

    • We may still observe local wrap-up effects in this case.
  • Mandatory commas may only be critical for comprehension in certain syntactic constructions, but not in others. In this case, we would observe the comma benefit effect only for some of the sentence types introduced.

Method

Participants

  • 32 undergraduate students (27 women) from the University of Valencia (mean age 21.97, range 19-39)
  • All native Spanish speakers
  • Normal vision and no previous diagnosis of reading disorders.

Materials and procedure

  • Materials
    • 130 sentences
      • 26 each with concessive, adversative, connective, enumerative, and incisive commas.
      • Two versions: one with the comma, one without it
  • Procedure:
    • Participants were asked to read the sentences silently.
    • Comprehension questions were asked after 26% of the sentences.
    • Participants were excluded if they answered fewer than 85% of questions incorrectly.
      • All participants scored above this criterion

Data analysis

  • We analysed the data by fitting Bayesian linear and generalized linear mixed models, using the brms package (Bürkner 2017, 2018).
  • Contrast coding:
    • Comma condition: present = -.5; absent = .5
    • Sentence type:
      • Treatment contrasts with adversative as the baseline
    • Fixed effects included the effect of the comma condition, the sentence type, and their interaction
  • Response times modeled using the ex-Gaussian distribution: The mean of the Gaussian component μ and the scale parameter β of the exponential component (equaling the inverse of the rate parameter λ) were both allowed to vary between conditions.
  • We allowed all possible random effects (comma condition and sentence type (including their interaction) by participant, comma condition by item)

Results

Global sentence reading time

Global reading time analyses

Global reading time analyses: beta

FFD

FFD analyses 1

  • Comma region: Comma effect (wrap-up) in the adversative condition, tends to be weaker in the other conditions
  • Nothing in beta

FFD analyses 2

  • Post-comma region: Wrap-up effects on the comma region in the adversative sentences are accompanied by faster processing in the post-comma region, again potentially weaker in the other conditions
  • Nothing in beta
  • No comma effects on the second post-comma region.

Gaze duration

GD analyses

  • Comma region: Not a lot of evidence for wrap-up effects in the adversative sentences, but might see them in mu for incisive and in beta for connective sentences
  • No comma effects in the post-comma regions (except for a slight effect in enumerations in beta)

Go-past time

Go-past time analyses

  • Comma region: again, wrap up effects in connective (mu and beta) and incisive sentences (just mu)
  • Post-comma region 1: Slight benefit for enumerative sentences (in beta)
  • Post-comma region 2: Slight benefit for enumerative sentences (mu and beta) and incisive sentences (in beta only)

Total viewing time

TVT analyses

  • Comma region: Wrap-up effects when comma was shown for connective (mu and beta) and incisive sentences (only mu)

TVT analyses 2

  • Post-comma region 1: Clear comma benefit (in mu) for adversative sentences, no evidence that effect is different in other sentence types
  • Post-comma region 2: No clear comma benefit except for a small benefit in enumerations (mu and beta).

Discussion

What have we learned?

  • Taking mandatory commas away does not seem to have the catastrophic effect on reading that one might expect.

  • In fact, while there are some clear wrap-up effects (and benefits from the processing that was already done on the subsequent words), overall the impact of commas seems to be quite minimal.

    • Impacts on processing seem to be local and rather short-lived.
  • Are commas really necessary? Can skilled readers live without them?

Future directions

  • The materials in this study were single line sentences.
  • Will commas be more useful when texts get more complicated?
  • Can we replicate our findings in other languages?
  • Are there any commas that are actually critically important?

Thank you

  • To my collaborators: Manuel Perea, Ismael Gutiérrez, and Ana Marcet

  • And to you for listening!

Results tables for FFD

FFD Comma region

FFD Post-comma region 1

FFD Post-comma region 2

Results tables for GD

GD Comma region

GD Post-comma region 1

GD Post-comma region 2

Results tables for Go-past time

GPT Comma region

GPT Post-comma region 1

GPT Post-comma region 2

Results tables for TVT

TVT Comma region

TVT Post-comma region 1

TVT Post-comma region 2

References

Chafe, W. (1988). Punctuation and the prosody of written language. Written Communication, 5(4), 395–426. https://doi.org/10.1177/0741088388005004001
Figueras Bates, C. (2015). Pragmática de la puntuación y nuevas tecnologías. Normas, 4(1), 135. https://doi.org/10.7203/Normas.4.4691
Frazier, L., & Rayner, K. (1982). Making and correcting errors during sentence comprehension: Eye movements in the analysis of structurally ambiguous sentences. Cognitive Psychology, 14(2), 178210.
Hirotani, M., Frazier, L., & Rayner, K. (2006). Punctuation and intonation effects on clause and sentence wrap-up: Evidence from eye movements. Journal of Memory and Language, 54(3), 425–443. https://doi.org/10.1016/j.jml.2005.12.001
Rayner, K., Kambe, G., & Duffy, S. A. (2000). The effect of clause wrap-up on eye movements during reading. The Quarterly Journal of Experimental Psychology Section A, 53(4), 1061–1080. https://doi.org/10.1080/713755934
Steinhauer, K. (2003). Electrophysiological correlates of prosody and punctuation. Brain and Language, 86(1), 142–164. https://doi.org/10.1016/S0093-934X(02)00542-4
Truss, L. (2003). Eats, shoots & leaves (First Printing edition). Gotham.